A Bayesian approach for identification of ice Ih, ice Ic, high density, and low density liquid water with a torsional order parameter

An order parameter is proposed to classify the local structures of liquid and solid water. The order parameter, which is calculated from the O-O-O-O dihedral angles, can distinguish ice Ih, ice Ic, high density, and low density liquid water. Three coloring schemes are proposed to visualize each of the coexisting phases in a system using the order parameter on the basis of Bayesian decision theory. The schemes are applied to a molecular dynamics trajectory in which ice nucleation occurs following spontaneous liquid-liquid separation in the deeply supercooled region as a demonstration.